The full agentic SDLC
Build with agents.
Keep the evidence.
Practical field guides for decisions, controls, and handoffs in agent-assisted software.
About the author
Marty Sampson
Marty writes practical field notes about building software with agents—grounded in clear decisions, bounded execution, and evidence another person can inspect.
Featured
Four useful starting points
Start with scope, source inspection, fallback attribution, or extension review. The library keeps every public guide discoverable.
RunbookRead →Field guideRead →Field guideRead →Field guideRead →
Bounded agent loop
Set scope, evidence, and stop conditions.
Inspect Omarchy source before install
Pin and hash official source text before crossing an installation boundary.
A fallback is another attempt
Retain attempt lineage, estimate provenance, and budget stops.
Agent extension review
Inspect manifest claims before installation.